Chris,
>Perhaps by studying talented programmers and 'scientifically' studying
>their cognitive capacities of various forms, we could begin to get a
>better picture of 'what' faculties make a these 'good' programmers. Of
>course, choosing these talented programmers is a whole other issues - and
>maybe the only way to do this would be through 'peer reviews'. And a
>study of 'cognitive capacities' may necessitate the evil of psychometrics.
What constitutes a talented programmer varies over different domain. The
include:
1) Ability to pump out the code (40K lines a year is top of the range)
2) Ability to create programs that will run in a constrained
environment (time/memory/cost)
3) Ability to work with incomplete and incorrect knowledge (no
documentation, out of data documentation)
4) Ability to take an existing program and modify/fix it (typical
applications these days usually come in at the 100K lines mark).
How do you tell the difference between somebody who has great application
domain knowledge and somebody who is great at writing software?
The trouble with peer reviews is that those making them rarely have access
to reliable measures of programmer performance (a general problem with
software development). So you could end up with good self publicists
and programers who once produced a great application and have been living
off it ever since.
Happy researching
derek
--
Derek M Jones tel: +44 (0)
1252 520 667
Knowledge Software
Ltd mailto:[EMAIL PROTECTED]
Applications Standards Conformance Testing http://www.knosof.co.uk
